Behavior recognition for Learning from Demonstration

Abstract

Abstract-Two methods for behavior recognition are presented and evaluated. Both methods are based on the dynamic temporal difference algorithm Predictive Sequence Learning (PSL) which has previously been proposed as a learning algorithm for robot control. One strength of the proposed recognition methods is that the model PSL builds to recognize behaviors is identical to that used for control, implying that the controller (inverse model) and the recognition algorithm (forward model) can be implemented as two aspects of the same model. The two proposed methods, PSLE-Comparison and PSLH-Comparison, are evaluated in a Learning from Demonstration setting, where each algorithm should recognize a known skill in a demonstration performed via teleoperation. PSLHComparison produced the smallest recognition error. The results indicate that PSLH-Comparison could be a suitable algorithm for integration in a hierarchical control system consistent with recent models of human perception and motor control.

Predictive Learning from Demonstration

“…In the present work, PSL is viewed purely as a controller and the forward model is consequently not considered. However, in work parallel to this, we show that PSL can also be used as an algorithm for behavior recognition [11], i.e., as a predictor of sensor values. A big advantage of using PSL for both control and behavior recognition is that the forward and inverse computations are in fact based on the same model, i.e., the PSL library.…”

“…Hierarchical representations following this decomposition have also been tested in an LFD setting [19] where the robot successfully learns sequences of actions from observation. In work parallel to this, we also investigates PSL as an algorithm for behavior recognition [11], exploring the possibilities to use PSL both as a forward and an inverse model. The present work should be seen as a further investigation of these theories applied to robots, with focus to learning with minimal bias.…”

Learning High-Level Behaviors From Demonstration Through Semantic Networks

“…The low-level controller learns and executes behaviors that are mappings of sensory-motor data to low-level actions (Billing et al, 2010a;Billing and Hellström, 2010b). In the presented work, the technique for learning is Predictive Sequential Learning (PSL) (Billing and Hellström, 2008).…”

A Formalism for Learning from Demonstration*

“…A key aspect of this approach is the pairing of forward models (predictors) and inverse models (controllers) in a model-free way. We are analyzing this issue deeper and propose a possible solution based on the algorithm Predictive Sequence Learning algorithm in other recent publications [16,17]. There are several approaches to identify relevant aspects of the task that do not employ behavior primitives.…”

“…Figure 7 illustrates a subspace of Π . With the ambiguities resolved, through human feedback or other kinds of bias, the resulting sequence represents an instance of π ∈ Π as defined in Equation 17. Various types of feedback from the human can be applied such that the ambiguous sequence collapses into a single well defined sequence of behavior primitives that will enable repetition of the demonstrated behavior according to the user' s intentions.…”